Image-processing apparatus, method, and computer program product for correcting skew in scanned images

Abstract

According to the present disclosure, edges of a document image contained in a read image read by an image-reading apparatus including a paper path are detected, a candidate edge-skew occurrence position is detected, a candidate content-skew occurrence position is detected, an accumulated-skew occurrence position is detected based on the candidate edge-skew occurrence position and the candidate content-skew occurrence position, skew of a document rear-part image rearward from the accumulated-skew occurrence position in the conveyance direction is corrected, and image data corrected for accumulated skew is acquired.

What is claimed is: 
     
       1. An image-processing apparatus comprising:
 an image acquiring unit configured to acquire read image data of a read image read by an image-reading apparatus including a paper path; 
 an edge detecting unit configured to detect edges of a document image contained in the read image and acquires approximate straight lines by extending the longest straight-line parts of the respective edges; 
 an edge-skew detecting unit configured to detect, as a candidate edge-skew occurrence position, on one of the approximate straight lines that is in a document conveyance direction, a position at which the one approximate straight line and an edge containing the longest straight-line part of the one approximate straight line start to separate from each other; 
 a content detecting unit configured to detect direction-identifying contents defined as contents usable for specifying directions within the document image; 
 a content-skew detecting unit configured to detect, as a candidate content-skew occurrence position, on the one of the approximate straight lines in the conveyance direction, a position at which a rear angle is larger than a front angle and at which the front angle first becomes equal to or more than a certain angle, the front angle being an angle between the one approximate straight line and each of the direction-identifying contents, the rear angle being an angle between the one approximate straight line and one of the direction-identifying contents that is located rearward from the each of the direction-identifying contents in the conveyance direction; 
 an accumulated-skew detecting unit configured to detect an accumulated-skew occurrence position, based on the candidate edge-skew occurrence position and the candidate content-skew occurrence position; and 
 an accumulated-skew correcting unit configured to acquire accumulated-skew-corrected image data by correcting skew in a document rear-part image located rearward from the accumulated-skew occurrence position in the conveyance direction. 
 
     
     
       2. The image-processing apparatus according to  claim 1 , wherein the content-skew detecting unit detects, as the candidate content-skew occurrence position, on the one of the approximate straight lines in the conveyance direction, a position at which a rear angle is larger than a front angle and at which the front angle first becomes equal to or more than a certain angle, the front angle being an angle between the one approximate straight line and a side of a rectangular graphic region, the rear angle being an angle between the one approximate straight line and one of the direction-identifying contents that is located rearward from the rectangular graphic region in the conveyance direction. 
     
     
       3. The image-processing apparatus according to  claim 1 , wherein the content-skew detecting unit detects, as the candidate content-skew occurrence position, on the one of the approximate straight lines in the conveyance direction, a position at which a rear angle is larger than a front angle and at which the front angle first becomes equal to or more than a certain angle, the front angle being an angle between the one approximate straight line and a ruled line, the rear angle being an angle between the one approximate straight line and one of the direction-identifying contents that is located rearward from the ruled line in the conveyance direction. 
     
     
       4. The image-processing apparatus according to  claim 1 , wherein the content-skew detecting unit detects, as the candidate content-skew occurrence position, on the one of the approximate straight lines in the conveyance direction, a position at which a rear angle is larger than a front angle and at which the front angle first becomes equal to or more than a certain angle, the front angle being an angle between the one approximate straight line and a character string, the rear angle being an angle between the one approximate straight line and one of the direction-identifying contents that is located rearward from the character string in the conveyance direction. 
     
     
       5. The image-processing apparatus according to  claim 4 , wherein the content detecting unit detects an angle between the one of the approximate straight lines that is in the conveyance direction and the character string in the document image by: detecting the character string; calculating areas of rectangles circumscribed to the character string while rotating the character string, the rectangles each being defined by line segments parallel to the one approximate straight line in the conveyance direction and horizontal line segments; and specifying a rotation angle corresponding to the smallest one of the areas. 
     
     
       6. The image-processing apparatus according to  claim 4 , wherein the content detecting unit detects an angle between the one of the approximate straight lines that is in the conveyance direction and the character string in the document image by: detecting the character string; performing optical character recognition on the character string while rotating the character string; and specifying a rotation angle at which accuracy of the character recognition is the highest. 
     
     
       7. The image-processing apparatus according to  claim 1 , wherein, on conditions that the candidate edge-skew occurrence position and the candidate content-skew occurrence position have been detected and that the distance between the detected positions is equal to or less than a certain distance, the accumulated-skew detecting unit detects, as the accumulated-skew occurrence position, one of the candidate positions that is located rearward from the other in the conveyance direction. 
     
     
       8. The image-processing apparatus according to  claim 1 , wherein, on conditions that the candidate edge-skew occurrence position has been detected by the edge-skew detecting unit and that the direction-identifying contents have not been detected by the content detecting unit, the accumulated-skew detecting unit detects the candidate edge-skew occurrence position as the accumulated-skew occurrence position. 
     
     
       9. The image-processing apparatus according to  claim 1 , wherein, on conditions that the candidate edge-skew occurrence position has been detected by the edge-skew detecting unit and that the candidate content-skew occurrence position has not been detected by the content-skew detecting unit, the accumulated-skew detecting unit detects the candidate edge-skew occurrence position as the accumulated-skew occurrence position. 
     
     
       10. The image-processing apparatus according to  claim 1 , wherein, on conditions that the candidate edge-skew occurrence position has not been detected by the edge-skew detecting unit and that the candidate content-skew occurrence position has been detected by the content-skew detecting unit, the accumulated-skew detecting unit detects the candidate content-skew occurrence position as the accumulated-skew occurrence position. 
     
     
       11. The image-processing apparatus according to  claim 1 , wherein the accumulated-skew correcting unit acquires the accumulated-skew-corrected image data by: assuming that the document rear-part image is a region bounded by two circular sectors; and correcting skew by bringing the arcs of the circular sectors onto the approximate straight lines that are in the conveyance direction. 
     
     
       12. The image-processing apparatus according to  claim 1 , further comprising:
 a front-end skew detecting unit configured to detect front-end skew defined as an inclination of the document image with respect to the read image by detecting inclinations of all of the approximate straight lines with respect to the read image; and 
 a front-end skew correcting unit configured to correct, based on the front-end skew, the document image to erect the document image. 
 
     
     
       13. The image-processing apparatus according to  claim 1 , wherein the accumulated-skew correcting unit acquires the accumulated-skew-corrected image data by: assuming that the document rear-part image is a region bounded by two circular sectors; and correcting skew by geometrically transforming the arcs of the circular sectors onto the approximate straight lines that are in the conveyance direction. 
     
     
       14. The image-processing apparatus according to  claim 1 , wherein the accumulated-skew correcting unit acquires the accumulated-skew-corrected image data by: assuming that the document rear-part image is a region bounded by two circular sectors; and correcting skew by geometrically transforming the arcs of the circular sectors, in a manner that extends the shorter one of the arcs, onto the approximate straight lines that are in the conveyance direction. 
     
     
       15. The image-processing apparatus according to  claim 1 , wherein the accumulated-skew correcting unit acquires the accumulated-skew-corrected image data by correcting skew by performing lens skew correction on the document rear-part image. 
     
     
       16. The image-processing apparatus according to  claim 1 , wherein the content detecting unit extracts straight-line elements of the contents by a Hough transform. 
     
     
       17. An image-processing method comprising:
 an image acquiring step of acquiring read image data of a read image read by an image-reading apparatus including a paper path; 
 an edge detecting step of detecting edges of a document image contained in the read image and acquiring approximate straight lines by extending the longest straight-line parts of the respective edges; 
 an edge-skew detecting step of detecting, as a candidate edge-skew occurrence position, on one of the approximate straight lines that is in a document conveyance direction, a position at which the one approximate straight line and an edge containing the longest straight-line part of the one approximate straight line start to separate from each other; 
 a content detecting step of detecting direction-identifying contents defined as contents usable for specifying directions within the document image; 
 a content-skew detecting step of detecting, as a candidate content-skew occurrence position, on the one of the approximate straight lines in the conveyance direction, a position at which a rear angle is larger than a front angle and at which the front angle first becomes equal to or more than a certain angle, the front angle being an angle between the one approximate straight line and each of the direction-identifying contents, the rear angle being an angle between the one approximate straight line and one of the direction-identifying contents that is located rearward from the each of the direction-identifying contents in the conveyance direction; 
 an accumulated-skew detecting step of detecting an accumulated-skew occurrence position, based on the candidate edge-skew occurrence position and the candidate content-skew occurrence position; and 
 an accumulated-skew correcting step of acquiring accumulated-skew-corrected image data by correcting skew in a document rear-part image located rearward from the accumulated-skew occurrence position in the conveyance direction. 
 
     
     
       18. A computer program product having a non-transitory tangible computer readable medium including programmed instructions for causing, when executed by a computer,
 the computer to perform an image-processing method comprising: 
 an image acquiring step of acquiring read image data of a read image read by an image-reading apparatus including a paper path; 
 an edge detecting step of detecting edges of a document image contained in the read image and acquiring approximate straight lines by extending the longest straight-line parts of the respective edges; 
 an edge-skew detecting step of detecting, as a candidate edge-skew occurrence position, on one of the approximate straight lines that is in a document conveyance direction, a position at which the one approximate straight line and an edge containing the longest straight-line part of the one approximate straight line start to separate from each other; 
 a content detecting step of detecting direction-identifying contents defined as contents usable for specifying directions within the document image; 
 a content-skew detecting step of detecting, as a candidate content-skew occurrence position, on the one of the approximate straight lines in the conveyance direction, a position at which a rear angle is larger than a front angle and at which the front angle first becomes equal to or more than a certain angle, the front angle being an angle between the one approximate straight line and each of the direction-identifying contents, the rear angle being an angle between the one approximate straight line and one of the direction-identifying contents that is located rearward from the each of the direction-identifying contents in the conveyance direction; 
 an accumulated-skew detecting step of detecting an accumulated-skew occurrence position, based on the candidate edge-skew occurrence position and the candidate content-skew occurrence position; and 
 an accumulated-skew correcting step of acquiring accumulated-skew-corrected image data by correcting skew in a document rear-part image located rearward from the accumulated-skew occurrence position in the conveyance direction.